Overview
I trained a multimodal AI model to predict prognosis for cancer patients and compared its performance against an NLP-only approach.
Why it matters
If the NLP approach performs as well as the other modalities, a model like this could be deployed worldwide in places where digital infrastructure, like whole-slide imaging, isn’t available.
Background
This research was part of my Dartmouth EDIT AI internship (June to August 2026), where I researched machine learning applications for cancer patients and quantified performance across modalities such as text and whole-slide images.